How the rankings are made
Every number has a source and a date
Each ranking names the public source it is built from, the exact measure, the time the data was fetched and, where the source reports one, the date the data is for. The source's terms are reviewed before a ranking goes live, and the source is credited on every page.
Scheduled fetches, immutable snapshots
Rankings refresh on a fixed schedule. Each refresh stores the raw response and produces a snapshot: the full ranked set with values at that moment. Snapshots are never edited or overwritten; the history chart and the yearly archives are built from them.
A quality gate before anything is published
Before a snapshot becomes the live ranking it is checked: enough rows, no unexplained collapse in the row count, no implausible swings across many items, no unknown items appearing in bulk and no source date going backwards. A snapshot that fails any check is held for a person to review instead of being published.
Corrections are visible
If a published snapshot turns out to be wrong, it is marked as superseded and a corrected snapshot is published with a note. The original is kept for the record.
What the rankings are not
They are measurements of one metric from one source, not judgements of quality and not advice. Each page states what its measure does not capture. Where a page shows affiliate links, they are labelled and never affect position or inclusion.
Ties, changes and shares
Items with equal values keep their previous order, then sort by name. The change column compares with the previous published snapshot. Where a measure is additive, each item's share of the top-10 total is shown.